How Bared Turns Trusted Data Into Decisions and Action | Triple Whale | #646

Fourteen years ago, Angus Growney realised he could game his own attribution model. He's spent every year since chasing numbers he can actually trust.

That hunt is what brought Angus Growney to Bared, the podiatrist-founded Melbourne footwear brand where he leads performance marketing across a business that now does roughly 70% of its trade online and is expanding into the US and New Zealand. Alongside him is Zach Rego, Chief Revenue Officer at Triple Whale, who joined in 2022 as it launched the first-party pixel built to stitch conflicting platform numbers into one customer journey. Zach now calls Triple Whale the AI operating system for ecommerce, with more than 60,000 brands on it. Nathan recently onboarded Triple Whale for a client himself, so he brings his own reactions to the platform along the way. This isn't a story about finally cracking attribution. It's about what happens once you trust your data, when the work shifts from understanding the numbers, to making the decisions, to letting AI take the action.

Today, we're discussing:

• Why trusting the data is harder than building the report, and the three months Bared spent in the warehouse before switching Triple Whale on.
• How Angus runs Australia, the US and New Zealand off one collated view, and the numbers he makes decisions on each week.
• What Triple Whale's first-party pixel and Moby actually do, explained plainly by Zach Rego. Inside True Classic running its entire Meta budget through Moby with no human in the loop, guided by a ten-page rules file.
• The PMax debate an AI settled, and why taking the opinion out of the room led to the right call.Where to draw the line on AI autonomy: Copilot, Autopilot and guardrails, and what to automate versus keep human.

Connect with Zach Rego | Angus Growney | Explore Triple Whale | Bared

Transcript

Copy

Cold Open And The Big Claim

SPEAKER_03 0:01

If you wait another six months to get started, you're like 10 years behind. Their entire meta budget is managed by Moby. No human in the loop. Just Moby.

SPEAKER_00 0:11

Immediate buying is dead, I think they're saying. So press the button and come back later.

SPEAKER_03 0:18

I'm Zach Rigo from Triple Whale, and in this episode of Add to Cart, I talk about building a trusted measurement foundation and using AI to turn that data into better decisions and real action.

SPEAKER_02 0:29

14 years ago, Angus Groni sat in a meeting about a global attribution model and realized that he could game it. He could buy media in a way that made his numbers look brilliant without actually selling an extra thing. He has spent every year since chasing the opposite of that data that he can actually trust. Today, Angus runs performance marketing at BERD, the podiatrist founded footwear brand that is expanding fast in the US and New Zealand. Alongside Angus, I've got Zach Rigo, Chief Revenue Officer at Triple Whale, joining us late in the day from the US. Zach came on board Triple Whale in 2022, right as they were launching their first party Pixel. This is the thing that is built to stitch all those conflicting platform numbers into one customer journey that they are now famous for. These days, he calls Triple Whale the AI operating system for e-commerce, with more than 60,000 e-commerce brands on it. I've actually just onboarded Triple Whale for a client myself. So I jumped into this conversation with my own reactions and input along the way. Now, this isn't a story about finally solving attribution. It's about what happens once you trust your data when the work shifts from understanding data to actually making decisions and then to letting AI take actions. Along the way, Angus and Zach get properly nerdy on how Triple Wales AI Moby works in the real world, from settling real team debates to running whole media budgets on its own and even briefing creative. This episode is really practical. It's honest about where we're at with data and the guardrails around it. And there are some really cracking examples in there, not only from Australia, but from around the world. Now, quickly before we get in, this is a special exclusive episode brought to you by our friends at Triple Whale. However, we have made sure that there is so much in this episode when it comes to measurement and data, whether you are using Triple Whale or not. A big thank you to that team for making it happen. And if you want the cheat sheet from this episode, you want the links and the shortcuts and everything else that we discuss, head on over to addecart.com.au and sign up for the newsletter. We'll send you the cheat sheet on the day that every episode is released so you never miss a thing, even if you cannot listen to the whole episode itself. All right, let's get into it. Here is Chief Revenue Officer Zach Rigo

Why Measurement Trust Comes First

SPEAKER_02 3:00

from Triple Whale and performance marketing lead Angus Groni from Bad. So, Zach, for people who may have heard of Triple Whale, and I'm sure there's not many people in e-commerce who haven't heard of Triple Whale, or maybe they've seen the stuffed blue whale on stage at some point and like, what is that? Can you explain what triple whale is and the problem that it's solving in e-commerce?

SPEAKER_03 3:21

Yeah, so triple whale is the e-commerce operating system for all really modern operators. And our original thesis was how do we bring all the data that an operator needs to look at into one place and make it real time? Business is running 24-7. It needs to be accessible, it needs to be quick to get insights from and take action from. That thesis has remained the same. And now instead of a dashboard that's kind of a no-code, you know, dashboard that are a dime a dozen today, we've not only added more data, but we've actually brought an AI layer on top of it that allows you to activate that data. And more data and insights, more platforms, more marketplaces is not easier to manage. So having AI that can look across the whole business to take action with you is a total game changer. And that's really what we've been focused on for the last couple of years.

SPEAKER_02 4:04

I remember when Triple Well first came on the scene and it was really groundbreaking at the time because everyone was managing ads in different platforms. And what Facebook might be telling you was very different to what Google was telling you. And the piece in market was around actually one source to give you the right attribution. It still feels like no one has solved attribution fully, but platforms like Triple Whale are the best source to help balance that out. Is that still a key part of the problem that you're solving with customers who are coming to you for the first time?

SPEAKER_03 4:38

100%. So we launched uh proprietary first party pixel in 2022. That was right when I joined. And the goal of that was really to stitch together the customer journeys that, you know, the ad platforms are all saying, hey, we did it. Oh, we did it. And when you look at that, you're kind of seeing a lot of conflict. So now the Pixel brings one customer journey, one order, you can see end to end every step they took, every platform they touched, all clicks on the website. Is it perfect? No, because people view things, right? And then they leave and make a purchase through Google or your email provider. We're getting closer. So we now have partnerships with a lot of the major ad platforms to get deterministic views matched back. So meaning if I viewed, you know, an ad on my mobile device and then came back later through a click and converted, we're now stitching together that journey with the ad platforms. We've also brought in other more advanced measurement models in addition to MTA with a full MMM and incrementality. And those three measurement methodologies now work together within our platform. So, you know, we've always said no one measurement methodology is correct. And we've now brought them all to the table to try to help our customers, you know, triangulate is the word we always heard on demos and in customer calls. And now we're kind of bringing the data together and the triangle together and giving you one source of truth. So I think we're getting, you know, dangerously close to answering that question.

SPEAKER_02 5:59

How do you approach

What Triple Whale Actually Does

SPEAKER_02 6:00

that? Obviously, you're not gonna give us your secret source to attribution, but I'm keen to understand when Facebook is saying this and Google is saying this, how do you play that triangulation role to go, actually, this is really what happened? Like you're gonna have to make some decisions or some algorithmic choices there.

SPEAKER_03 6:17

Yeah, it's funny. You know, we battle with, and I love Angus's perspective on this, we battle with brands sometimes on this where if we agree with their hypothesis, it doesn't matter if we're right or wrong, if we agree with the hypothesis they came to us to get a stamp of approval from, then we're right. And if we disagree with their hypothesis, we actually have to challenge the customer to make a change based on a different insight that we've provided. And I spend a lot of my time talking to customers that are, you know, some of our more advanced, larger customers, where I'm like, can we do this test together and like actually make the change that we're saying that you don't feel comfortable with, but let's do it. And that's been fun because we are right a lot of the time, not always for sure, but it's interesting to see that dynamic of a lot of people want to confirm their expectation of what measurement should be or what the platforms that are performing best are because they've had that internal debate. And uh when we do that, it's easy. And when we don't, there is a bit of pushing and pulling. Incrementality studies, there's no, you know, BS in it. It's you do the test, you see what the lift was of that channel. And that actually feeds our MMM and feeds some of the MTA so that you can now, you know, have a model that's really bespoke to your business and also, you know, change how you operationalize your daily budget changes through our, you know, MTA and Pixel data.

SPEAKER_02 7:30

Okay. So, Angus, this is a perfect throw point. Zach, thank you for for doing that. Angus, you're managing the performance marketing at Bed, and we were lucky enough to have Alex from Bed on almost two years ago now, I think. So amazing business, a footwear business, if you haven't seen them, expanded internationally over the last couple of years. Great, great Australian business. What were the internal conversations for you that led you to almost having a mediator to give you better insights into what's actually working?

SPEAKER_00 8:00

You know, it's so funny. Zach, we should have hung out before this because we could have talked for hours on this. It's so interesting to hear that even the same conversations Zach is having near supplies, you have those internally, right? And whether you use triple whale or not, it's like, where's the source of truth? This has to be 100% accurate. And we just have to realize that that isn't true and that like the testing and the proof is the sales at the end. That's what we track and measure. You need one single source of truth to kind of guide that. And triple whale has been that since 2024 when we first came on. You know, my attribution journey actually started 14 years ago with I think I could say this, with HelloFresh when they first came to Australia. And I had never even heard of, you know, attribution, MMM, these kind of things, what, you know, traditional media buyers. And I was buying TV for them here in Australia, you know, we're having meetings out of Germany about their in-built attribution model, and I was like, what is this? You know, like is this what global businesses are doing? Because I hadn't heard of people in Australia, like at that level, from a traditional media buyer, and then also even just in the digital landscape. But one thing I did find with that is I could manipulate their MML, right? Like the way I bought media, I could do things that would increase our performance from their point of view that wouldn't actually generate revenue, right? And so that's led me down this path over the journey of like, okay, well, if we can manipulate it, how accurate is it? The data inputs and how important they are to be accurate before we make a decision. And when I first came to bed, the first thing I did was like, you know, we're spending enough, we're growing quick enough, we need attribution software from a performance team to guide the rest of the business. You know, we use some other suppliers, but again, it's like the data coming in, do we trust it? And then back to the main point is what, you know, senior leadership is like, you know, why is that not 100% correct? You know, or why are you questioning the data in there? And it's like that journey for everyone to come along that, you know, the viewable results aren't being tracked back then. You know, there's more to a customer's journey. You know, they might have talked to a friend, then they might have had another 11 touch points before they've come into the picture. And I think, as we've seen over the last 12 months at Triple L, that gap has closed. You know, while media channels, Meta, Google, they're going the opposite way, I think at the moment, the model conversion elements are increasing from what I'm seeing. There's more leaning into viewable results and clickable results. And so I don't think I've ever been more confident in like putting data back to the business. And then I think as Zach had said before, actually being able to track it better with triple whale than we have ever before.

SPEAKER_02 10:39

How much of your time is spent with triple whale or even any of your data in justifying your results versus making decisions from them?

SPEAKER_00 10:48

Making decisions is probably the bigger thing now because every single thing that we set up, if we might even just set up a campaign because we think it's going to work, we now go into an agent or our Moby uh media strategist and we write out our reasoning for why we're setting up a campaign, what we've done with the ad groups, what creative elements we're putting in there and what we expect, you know, as the critical thinker, the outcome should be, right? Yeah. We're then getting the agent to go circle back to me, depending on the spend, of course. It's like circle back to me every three days. Tell me if it's heading in the direction I think. And also tell me if it's not and why. Give me your perception, right? So, you know, you used to write these down or you do an A-B test, and then two weeks later you're like, oh, does that work? Or have we actually checked in on that? No, we're getting emails in our inbox in our Slack going, hey, you need to check this campaign, or it's going really well. We can actually scale that further, or it's coming back with creative recommendations that go to a creative team to be like, hey, we need different creative, or this one creative's working, but the other seven aren't. Can we make an iteration of that? That makes a lot of sense.

SPEAKER_02 11:53

And that was the situation that I was in with the client that we're we're working with right now. It wasn't so much around attribution, even though that's important to get one source of the truth or one source of data. It was about how do we enable the team to make decisions quickly? Because this was a case where we would bring performance media buying in-house, and we needed to enable them to make quick decisions, almost like a trading, day trader based on what they're seeing. Zach, are you seeing the use cases for how people are using data out of triple whale change and develop?

SPEAKER_03 12:28

100%. I'll give a couple of good examples. I think this made me think of a good one, which was the internal debate. Whether you're using triple whale or not, get your data organized. When that internal debate happens, let the AI be the tiebreaker, whether it's us or not, because it's crazy the level of depth it can go to. And we have a great example of this with one of our customers who had an internal debate about PMAX campaigns. And Moby was like, kill it and kill this specific slice of it. And of course, it agreed to with the director and the media buyer was on the other side and they killed it, and it

Triangulating Attribution With Three Models

SPEAKER_03 13:00

was a great result. So I think it's like those internal debates, when you're having them, we've never been in a better moment to be like, you know what, let's take opinions out of it, which we all have, and just let the data tell us something and take that action and see what happens. And I feel like that's really uh empowering and freeing and allows the decisions to flow faster. There's another story with a customer of ours called Origin. So we have access to not only all of the platform data for media buy, but also social comments. So all of your comments on your social, all of your review data. We know which products are returned. We had a customer who actually changed their product in scene by a half of an inch on a pair of jeans that has made them an incremental $100,000 on that product because they were getting so many returns because the sizing was just ever so slightly off. And the only reason they figured that out is because it wasn't like if you looked at one source, your social comments, it was like a problem. Or if you looked at one other, it's when you bring it all together and you let, you know, AI go through it that you're like, whoa, we got a problem here. They ship it to the product team, they fix it. It's like that wasn't our original use case. But we didn't think of that. The customer thought of it and it's incredible, which is really fun. Like I love hearing those customer stories because it's just like, oh, we're not demoing that, we're not talking about that. But that is just like an amazing outcome for the brand and changes their business and makes their lives easier. So that's really fun.

SPEAKER_02 14:14

Angus, have you had any of those breakthrough moments where it's like you're pulling all this data together, even though I can imagine you seem like someone who's across the data in all the platforms. But when you see it all in one place, have you had any of those oh shit moments?

SPEAKER_00 14:26

I can give you like one that was almost also Triple Whale's little sales pitch to get us on and originally. And you know, it's not that anyone else is tracking is bad, but data is missed, right? And so, like one of the first things we set up and the team at Triple Whale were like, you know, don't worry about the costs, we'll be able to save it in a flow. And so basically there's abandoned checkout emails, you know, not the clavios, we love clavio, I love it, but some of their pixel data is missing, right? And so the all the abandoned checkout flows, we then set up a backup with Triple Wales Pixel. So anyone that didn't get picked up in Clavio's flow was then picked up in a flow by the triple whale pixel. And it was picking up about 20% that was missed, right? So that ended up being a couple of thousand dollars a week that we were now recapturing just out of a flow because of missed data. And, you know, obviously after two weeks, I was like, well, the platform's paying for itself after one, you know, our first kind of test. So that was a really good one. But like we started this pump there from Zach in the background.

SPEAKER_01 15:24

The sales team didn't lie. Good for them. No, yeah.

SPEAKER_03 15:27

We do tell the truth, I promise.

SPEAKER_00 15:29

Most of But you also you don't know what you don't know, right? Like, and we understand that data's missing, and you're just trying to get the best out of every system. And so, like, there's a lot of those small little things that we've picked up along the way that we're like, oh my God, how did we not see that earlier?

SPEAKER_02 15:46

You mentioned before you dropped the the word trust, being able to trust the data. And I think regardless of whether you're doing it on an Excel spreadsheet, whether you're looking in platform, whether you're using a platform like Triple Wale, as a performance manager, you need to trust that data. Like if you spend your time second guessing the data and trying to verify whether that data is actually real, you're wasting so much time. Angus, what are you putting in place or what are your guardrails to make sure that you trust your data?

SPEAKER_00 16:16

I think we spent three months getting the data right first with the triple well team, you know? Before we even got going, it was let's get it all into a data warehouse, let's make sure it's all in the tables that we need, and then let's justify it over a couple of months to make sure everything's lining up perfect before we even press go on triple L. And I think that time spent with the team over there, you know, there was no Australian team at that time. So there were some early and late nights with the devs over there. It gave us and the team so much more confidence in what we were looking at because we had spent six to eight weeks with the team, line by line, going through all the bits of data, ensuring that, okay, when we move forward and we make marketing and business decisions off the back of it, we can trust that it's correct.

SPEAKER_02 16:59

So are you connecting directly your platforms to Triple Well or are you passing through a data warehouse?

SPEAKER_00 17:04

A bit of both. Okay. So performance channels are all kind of directly in there, and then we've also got a big query data lake that connects into.

SPEAKER_02 17:12

Okay. And Zach, from your perspective, I know yes, it's easy to connect into your Shopify's, your Clavios, your Metas. We did that in less than an hour. That was easily done. And there is a deeper part of Triple Whale, especially when it comes to business data and pricing and margins. How deep do you see most of your customers go in connecting that data up?

SPEAKER_03 17:36

Yeah, it's it's dependent upon the size and scope of the brand. Uh, you know, the SMBs are, they almost self-onboard most of the time. So I've we say brands doing, you know, zero to 10 million where your tech stack is, right in our wheelhouse, you know, Shopify, Meta, Google, maybe TikTok, you know, Clavio, and you're kind of plugged in and ready to go. Implementation for those is an hour, honestly. Mid-market brands and upper market brands, they'll bring in their full operating expenses. You know, some brands that are literally optimizing on our Pixel page down to contribution margin per order. You know, and that's a custom column you can create. So, you know, every transaction that happens, they see, oh, we made $15, we made $20. And then they kill or crank up and scale campaigns and ad sets and ads based on that column. So you can get incredibly nuanced, and that's sometimes integrating your data warehouse, like Barrett has done. You can do a Google Sheets kind of automated import, and we get all the way down into your operating expenses. We have folks that are like, my rent's paid on the first, and I'm gonna upload that. And, you know, I pay this bill on this and I'm gonna upload that, and it all gets calculated in. And that's really, you know, you can log into triple on your phone and see your profit for the day. And there's a lot of value in that for us because people keep logging in. They like that dopamine hit. Yeah. And also great for operators that have, you know, one place to look for how they're doing on a daily basis.

SPEAKER_02 18:55

That is what surprised me the most. Well, not the most, but I went in and had a look straight away after connecting everything up. And I was like, advertising agency costs in there. We can put all these other costs to get a total marketing efficiency ratio in here that's not just performance related and have a look at total marketing costs, which I didn't even talk, honestly, to the sales guys about that beforehand. But it was nice to see it because I really like the MER measurement. On measurement, can I ask you, Angus, what are the key measurements that you're using in triple well? What is like the first thing that you open up when you get in on a Monday morning? What's the triple well dashboard that you're looking at?

SPEAKER_00 19:30

This is a very interesting question because over the last three months, so many wild things have happened in the platform that is now changing. Do I open my email or Slack first where it's talking to me? Or do I open up the, you know, the attribution uh overview that I normally have looked at for the last few years, right? So like the attribution is still at the core of what we are using it for, right? Obviously, yes, there's a lot of other exceptional things coming off the back, but we're always looking at attribution, campaign performance, creative performance.

How Bad Data Breaks Decisions

SPEAKER_00 20:00

Now, that has slightly changed because your question was, what am I opening up first? We have now built, you know, reporting agents, strategists, creative strategists as well. And so they're developing daily reports, weekly reports, monthly reports with all the core metrics that we look at as a business and sending that to all the performance team, the creative needs to go to the creative team of what creative worked last week, what didn't, what could we engineer this week from a new creative perspective? And then anything from a senior management, like what are they looking at? And that's where it's back. We have plugged in everything to triple well. So it is like, how did the stores go? What do we spend that drove any traffic into those? It's like this overview of the entire business that it can provide on a Monday morning email is amazing at the moment. Do you think dashboards and reports are on the way out? Absolutely. You know, I think the more that we're starting to integrate it even just into Slack, going into Moby, talking to it, you know, internal performance teams are always very small, right? Like generally it's one or two. There could be a couple more, but there's not a whole lot. And it's pretty cool for those teams to be able to query things in real language. Yeah. You know, it's like, you know, you always have those brain thoughts, brain thoughts, and you're just like, I wonder what happened there. And it's like, great. I said set up a Mobi, I'm asking that, open up a new chat. I've got this question. Most of the reports have got your basics in there. But now it's starting to build out that really deep thinking of like what happened in this state in the US and what were those customer journeys over the last six months. You know, why did we have this big week in, you know, Colorado? And so without me having to spend a day figuring out like what happened in the state, what ads we ran, what happened in Google ads, it's like five minutes later, I've got this beautiful report that I'm like shared with the rest of the team. And I think the shareability of some of those reports for everyone else to understand has been such a game changer for me because sometimes we become too nerdy in yeah, this is the metric, and here's the data, and everyone's like, I don't really know what that means. You know, and so But it justifies our jobs. Right. And we think it sounds smart, but it's like the more you hand over to Moby and just AI in general to explain what's happening is great for the wider team. Yeah, that makes a lot of sense.

SPEAKER_02 22:12

Zach, I think you'll love this. When we implemented Triple Whale and we opened it up straight away, gave the team access. The first thing that comes up isn't a dashboard or even numbers, it's Moby. So the first screen you get is Moby. Like, what do you want to know? And I was expecting a bit of resistance, especially from teams that have been used to the same dashboards for years, like looking at the same metrics the same way for ages. But the natural behavior that I observed was them just typing in a query that was top of mind and getting answers and going, holy moly. Like they weren't worried about losing dashboard straight away. They'll we'll definitely build dashboards down the track, but they're actually saying, Oh, I'm not as worried about it. I'm not rushing to get to my dashboards and perfection. I'm actually typing my queries in straight away. Are you seeing that change with all your clients?

SPEAKER_03 23:03

Yeah, I think it's funny. You know, Angus brought up a really great point. We used to look a lot at like daily active users. Who's logging in? How often are they logging in? And going back to like the summary page on your mobile phone, like that was, you know, one of our number one daily active user hits was just mobile opening up the app. Now we're looking way more at are you connected to Slack and do you have an automation running? Because listen, if AI can do anything great today, it's limit clicks, right? I don't have to click in. I don't have to open up this tab. I don't have to drill into this campaign, this ad set, this ad to get this one insight to then go on with the rest of my day. It's I can wake up and that insight is delivered to my inbox at the time that I would have done 10 clicks. So I think it's interesting, like what we look at as performance of a good customer has evolved a ton. And it's right, it's how many chats are they doing, how many automations have they set up? Do they have Slack connected? Slack is a harness for Moby. Like it learns the context of the Slack channels you added in. So you can actually not only ask questions about the data in triple whale, but you can ask questions about the argument that Angus and the other person had three days ago and how the data's evolved based on that argument to see who won those debates that we have. It's really so fun. And I think we don't even quite know the limits of it yet. We use it internally. So we use triple whale internally for all of our BI and reporting and analytics. And it's been uh it's been really wild to see what people have been able to build with it.

SPEAKER_02 24:23

I feel like you need like a Mobi almost like in on an Alexa, but you need a Mobi on a desk within Teams just to settle arguments. It's like, all right, we've argued about this for five minutes. Just send it over to the triple while Alexa and tell us what the real answer is.

SPEAKER_03 24:37

So there is a microphone button in Mobi. You can have it transcribe your entire conversation and then hit send and have him answer it for you. So it's it we've done that. We've done that in our own executive meetings. It's been fun.

SPEAKER_02 24:49

That's great. So tell me about Moby. We recently got Moby 2, which was released a couple of months ago. Tell me about the big upgrade, because it is a big upgrade from Mobi to Mobi 2. What's changed and what are the AI capabilities now to take that data that is in the platforms and really utilize it for e-com teams?

SPEAKER_03 25:08

Yeah, so a few big changes there. Moby one was great. I think for the time and place, it was as good as you could get. And it was basically ask a question, get an answer. It could write SQL, so it could create dashboards or tables for you, but it wasn't doing work for you. So it was very similar to the the old way of working. It was like, I need an insight, I'm gonna go to a dashboard. Well, if I don't have that dashboard, now I can ask AI and it can create that dashboard and give me some basic insights. Moby 2 is now agents with subagents. The cool thing is as the labs and the frontier models get better, which we just released 5.6 today and we tested it for open AI, our product gets better. But Moby now takes action for you. So we have brands, you know, that are True Classics is managing their entire meta budget. And they've said this publicly, so I'm hoping to say it. Their entire meta budget is managed by Mobi, no human in the loop.

SPEAKER_02 25:58

Just Mobi.

SPEAKER_03 25:59

Making decisions on spend like budget changes, thousands of budget changes a month, turning campaigns' ads on off. Now, you know, we've got guardrails and they put in a markdown file that's 10 pages long of every decisioning process that they would make. But media buying is mostly a decisioning making process, right? I look at the campaign level, I see this thing, I drill down to the ad set level, I see this thing, I drill down to the ad level, and I make decisions off of that. And if you feed Moby all of those decisions and then you tell him how often you want him to review it, he can go and take action for you across Meta, across Google, across channel budget allocation. It's really wild. We have a lot of customers that have taken that leap. And it's been fun to see them kind of crawl, walk, run. And, you know, True Classics is sprinting, but it's been really, really fun to see people take action with their media buying process through AI.

SPEAKER_02 26:49

I can imagine their budget isn't small.

SPEAKER_03 26:51

It's a lot. I won't say that publicly, but it's a lot.

SPEAKER_02 26:54

And I think that's fascinating because we originally started talking out around moving with analytics from understanding

Agents Replace Reports And Speed Up Teams

SPEAKER_02 27:00

the data, you know, that attribution, which to your point, Angus is like a 14-year-old problem, if not more, into being having the teams making decisions. And now we're actually saying it's not just about decisions, it's about automating and taking action. So it's a big shift from just understanding data. From your perspective, Angus, are you letting go anymore and allowing more automation based on the data that you're getting through?

SPEAKER_00 27:27

That's a great question. I think as media buyers, we don't like letting go. But this year, absolutely more than ever before. I think everything that comes out of triple well, we've basically been on the beta as well. So we've been at the forefront of testing. We're the first in Australia to test the A and B, which is the autonomous media buyer in there. You know, and that was probably the first step to going, okay, and I think perfectly, as Zach said, you have to be able to engineer your own skills into it for it to work really well. You know, I think we think about everyone connecting, you know, Claude to the Meta MCP, and it's like it doesn't have the context. You know, it doesn't understand the customer journey, doesn't understand the product and the profit and where the business is heading. And so I always worry about people who are going down that road when there's a solution here that you can plug in everything, you can build your own skills, it's got the attribution modeling over the top. And we build the A and B because we were like, how are we going to scale in the USA when we are asleep when it's daytime? We need someone to make those decisions for us, and we're not looking to hire a USA team, right? Or even an agency at that point. So we don't know the state level trends and data. So what we have been doing is like not even just like East Coast versus West Coast. It's like, well, what's happening in Chicago versus New York, looking at Florida, and we're trying to go, well, what shoes are relevant at times over the year? And we're uploading that intel into the system to try and predict, like, when should we be shifting from sandals to boots? You know, like when do we shift from flats or flats is actually growing? So how do we push that and scale it further? And so it's really about like challenging your own thinking to be like, one, is that right? Build a skill for it for the A and B, or now there's media strategists built into different sections and go, well, what if I just let it do that? You know, like let it run for a test, set it up to track the test, predict the outcome, and then pay attention to it, right? And it's actually amazing to watch this evolve. And like, I think the last two months has been the most exciting time I've had in this role. It's allowing you to scale internationally without the actual headcount and the speed that we can do things, turn ads off. You know, we're saying, you know, we build what we determine is a good ad, right? If it hits, you know, three times spend, no conversions, turn it off. If it's struggling for spend because other things are taking priority, pick it up, move it into our secondary test account so it gets some spend to see if it does have legs, if it doesn't turn it off. Those are simple decisions the AI can make. We don't need to make those. We don't need to go into every ad account, turn off every, you know, turn off this ad, look at the data. It's like, no, no, if we're sleeping and the US and we need to turn ads off or move it into our scale campaign, then like let the A and B do that.

SPEAKER_02 30:09

I think that's really interesting because you're not talking about a performance marketing tool here. You're talking about a business analytics tool. Not to hype it up too much, but I can imagine that some of that data that you're getting out would be really interesting to your buying team, maybe even your retail team, your finance team, as they're making those decisions around inventory, even store locations as you're expanding internationally. How are you managing and giving access to that data to the broader teams who might not be marketing natives?

SPEAKER_00 30:38

So the great thing at bed, like there is no shut off to anyone. So anyone who wants to have access or we think should have access, it's like you're already at it and you could tee up a time with me to go run through what you need to get out of it. Now, before Moby 2, it was a little bit more difficult for teams. You know, I as Zach said, when you were chatting to it, it was more to build dashboards or come back with quicker questions. The deeper level of insight we probably weren't getting as much from. But since Moby 2, you know, I've run product teams through it because they're looking at, okay, well, what customers are buying that specific shoe? What's the time between them and buying this next shoe and what are they more likely to buy? So we're trying to build out like just that customer intelligence around who buys what, you know, and so then what shoes should we be developing for the existing customer in each persona? And like that's now providing insight for product teams. We have, you know, stock level alerts to be like, hey, we're going to sell out of that shoe if we keep it in ads. We need another shoe, product team. What should we be pushing in the US, in New York? And so like it's allowing other teams to actually start to get involved in what product should we be pushing, what product should we be creating? You know, and that's wider. Obviously, the creative team have a million different ways to use the creative elements. And we're very, very brand protective. And so, you know, AI is not quite there. But the difference that we've found with Moby currently is while AI can develop, you know, the image and go, this is what we should be running, Mobi's giving context on why, right? And so the context on why has been probably the most insightful thing for us because we're taking that and the image, putting that with the creative team and saying, don't worry too much about the image, you know, it's off-brand. But just think about the way it's describing why we would go down this road. And can we turn this into our own on-brand ad that uses the insights from Mobi? And I think that's been a bit of a change rather than just, you know, AI creative. It's actually, well, I think this ad would work because of the look and feel. It's autumn, you know, it suits these states in the US or here in Australia. And the creative teams actually enjoying that insight for them to then go create off the back of.

SPEAKER_02 32:48

So you're using it as an idea generator, essentially. Yeah. Are you seeing that, Zach? Because it's something that's not talked about a lot with Triple Whale is the creative side of it. We talk a lot about the data and the analytics side. How deep are people going in creative performance?

SPEAKER_03 33:03

Yeah. So two things that Angus said that I want to double down on. One is I said crawl, walk, run. I think the creative one's a really great example. Like we had very few customers that were just like, yeah, create images, you know, let it rip. A lot of them started with creative briefs, which is AI's like very good at looking at all the data we have, which is hook rate, you know, video view-through rate, all sorts of click-through rates, obviously. We know the copy, we know the image, we know we have computer vision to watch the video, listen to the music. It does a really nice job of organizing that and giving you an amazing creative brief. That to me is kind of the walk version of using creative in triple whale. But to be frank, I think last month we had something like 40,000 images generated in Triple Wheel. We have brands that redid their entire product photography on their website using a couple thousand SKUs. But we have access to all of the product imagery, every ad you've ever deployed. And that gives you a ton of flexibility. We've also built a Creative Studio

Moby 2 Takes Actions In Ad Accounts

SPEAKER_03 34:00

into the AI. So, what does that do? It basically gives you like a Canva style editor where you can select a snippet or basically pixels on an image that you want to change. So a good example of that would be like I took a model shoot, the color, but they're wearing the black shirt. And I want to change the black shirt to the white shirt. Well, Triple Wheel's already got my product imagery for the white shirt, and I can just pick that snippet, chat with it, change it, and go, overlay some copy, whatever I want to do, and push it. So we're seeing a lot of that kind of image manipulation, not, you know, make me an image of a pair of shoes that look like ours and, you know, let it rip with a model. That's still got a little bit of ways to go. We do have all the latest frontier models, and like some of the video models that TikTok just released and others are crazy. But still, for you know, a lot of brands that have a lot of you know high fidelity and hold their brand in close regard, I would not recommend that. But the briefing is an amazing tool for them.

SPEAKER_02 34:51

And I've got to ask you, what's powering Moby2 underneath? Because as you're going into creative generation, analytics, everything, there's a lot of different AI capabilities in here. What's powering the AI?

SPEAKER_03 35:04

Yeah, so we have access to all the latest frontier models. So right now, kind of the core model today is GPT 5.6, which launched today from OpenAI. We also have anthropic fables in there. For creative, it goes between for image generation, nano banana pro and has been the best, but we have access to Nano Banana 2 Pro, GPT Image, Grok Imagine, and Dola C Dream 5.0. And then for video generation, Vio, Gemini, Seed Dance, and Grok Imagine Video are all available. We have different defaults for each of those, Nano Banana Pro and Vio 3.0, but Seed Dance is actually, I think, will start to become the default on video over time. So we we put them all in, we let brands choose, we do default based on certain tasks. And Moby will actually iterate through certain models if they're not working well. So an example would be you want to create an image of something. If it tries nano banana two first and it doesn't work, it will then actually go and try nano banana pro to get the results that you're looking for. It actually happened to me today when I was trying to create a GIF of an image, which is really cool to watch. Takes a second longer, but it's pretty amazing to let the AI kind of figure out how to get the end result you're looking for.

SPEAKER_02 36:16

I can imagine that's pretty cool for you, Angus, if you want to experiment with different AI tools, but you don't want to sign up to a thousand of them.

SPEAKER_00 36:23

Yeah, absolutely. And, you know, I feel like we're only scratching the surface still with a lot of that too. You know, I'd love to hear, Zach, do you have any other examples of how people are using video? Because it's probably something that we haven't touched so much at bed. You know, we have in-house videographers and they have their own systems. They're probably the the one that's left that I haven't really got into the system. So any examples that have video would be great.

SPEAKER_03 36:46

Yeah, the best ones that we've seen are, again, going back to like a model shoot that you paid for. And you get a brief from Moby, let's say, that's like, here's this hook, this music works, this angle works, and you know you've got an image that similar, maybe different product, but you didn't do a video shoot for. Feed that image in and say, reference this video or this brief that I've got here, and it will actually take the video or the image, the brief from the video that worked well, and animate it in a similar style. And it's pretty darn good. Like these video models are getting very good. If you have, again, I've got a good video with the kind of the elements of what I need. I've got an example video and a brief. The one other note is like prompting AI to write the prompt for you will always work better than you prompting AI. So like ask Moby. Like, if you wanted to recreate this video with yourself, write a prompt to do that. And here's all the angles and hooks and things I'm looking for. And it'll give you the great prompt that you can go and let rip, and it'll be, it'll be quite good. You should test that. I'd be interested to see the outputs.

SPEAKER_02 37:45

Just don't show your uh brand team yet.

SPEAKER_03 37:47

Yeah. Deploy it without say Moby deployed this on accident, and when the results are amazing, you get to loosen the reins a little bit. Come on.

SPEAKER_02 37:55

Well, on that control, Zach, I'm really keen. I'm gonna put my risk hat on here. And obviously, we're connecting to a lot of data, especially when you're talking CRM and Shopify data. There is sensitive data in there that you want to connect to get the best results or to make the most value out of the connection. And then you are passing it out to obviously a lot of AI platforms. But then, like we said, the data is actually not limited to the platform now. We're, you know, to get best use out of it, we want to pass it to Slack. We want to integrate it into our other platforms that we're using. So all of a sudden we're bringing all this data in and then it's going out and it's obviously being manipulated and used by different AI platforms as well, and then out to different channels. What are the safeguards in place to make sure that that data remains secure?

SPEAKER_03 38:40

Yeah, so everything stays in triple well. Now, obviously, it is sending a Slack with plain text of, you know, the outputs that you're asking for. And, you know, the AIs are, the labs are reading it. You know, we've got as many safeguards as possible. You know, we have to go through pen tests with some very large enterprise clients and password flying colors. And, you know, we do have restrictions where we can turn off AI if we have brands that come to us and don't want anything going out to any of the labs, their teams are not allowed to use anything but, you know, co-pilot because they're a Microsoft shop and everyone's locked on Internet Explorer, which does happen.

SPEAKER_02 39:12

Oh, yeah.

SPEAKER_03 39:13

We can put those safeguards in place. So we give our brands and customers a lot of flexibility there. You know, the one thing I will say is if you're a brand trying to lock it down, there is no doubt one of your employees somewhere is using AI. So it's better to put rules around the AI than try to limit AI altogether. And, you know, we've had to do that internally. All of our teams use Mobi. We have access to all the models, but all of our data stays in Mobi and we use Mobi for everything. But it's an interesting challenge that is very different based on the size and scale of the brands that we're hitting and feeling and building the safeguards internally for as well.

SPEAKER_02 39:45

Yeah, I could imagine it gets harder as the companies get bigger that come onto Triple Wild.

SPEAKER_03 39:49

Again, the brands that win the next, and Angus said like the last two months have been the most fun. I think like the innovation the last two months is insane. Two years from now, we don't even know what that is going to look like. And if you wait another six months to get started, you're like 10 years behind. Like it's gonna feel that level of FOMO will be there. And we hear that a lot. I think it's gonna be hard to compete without bringing this into your workflow in some capacity.

SPEAKER_02 40:11

Yeah. And on that, Angus, I'm keen to hear from you. While we've got Zach in the room together, based on your use of triple A at the moment and knowing how fast everything is changing, where do you see you get more value out of Triple A? Like if you had to put a feature request in, in the vision of how powerful you see this could be on top of what you're using it for today, where do you think this could go?

SPEAKER_00 40:34

It's funny you asked that because when we were building the A and B for us, you know, I sat in a meeting every week with a team of 10 that were on the call and just pour me from my side. You know, and they were always asking that kind of question, you know, like, what is it that we want to get out of this? And like at the moment it's speed. You know, the speed to bring everything to the forefront means that we make quicker decisions as a brand. And like at the moment, that's winning, that's allowing us to scale like

Creative Briefs And AI Production Workflows

SPEAKER_00 41:00

internationally, not just US, but UK, Singapore, New Zealand, like without any headcount. So the critical thinking that we used to spend our 100% of our day on, trying to figure out things, it's now we're actually executing with 80% of our time rather than quite the opposite. The things that we're really trying to put into it is a bit more of that predictive modeling. And so what we're trying to get out of it is we know seasonal trends, right? Like as a business, we know when heels are in demand, we know when boots are at their peak, we know when the markets is starting to look like we can actually start to predict sales pretty closely by category anyway, but now we're putting that into Moby, right? And we're going, hey Moby, with everything that you know, what ads should we be creating, predicting over the next two weeks? And I think that's starting to get ahead of like the whole entire market because no one, everyone else is looking at behind data, right? They're looking at, well, this is what happened, so let's make a change. We're starting to get to the point of, well, we know the season's changing in the next two weeks. We know that on previous data and trends, these are the shoes that have worked. We know that these are the shoes that are coming out for us as a business. Like Moby can read a marketing calendar, products that are coming up. So we're starting to go, well, can we use that to go? This is the best time to release that color. That is the best time to release that kind of shoe. And so, like as we evolve in this space, it's like we're now working ahead rather than behind. And I think that's going to be the coolest thing that we can get into, you know, outside of the creative elements and everything. But for me, it's that predicting ahead and moving faster than the market.

SPEAKER_02 42:33

Yeah, that makes a lot of sense. So, Zach, I'll throw it to you. The roadmap, you mentioned how fast things are changing. Are you working on a three-month roadmap, six-month roadmap, 12-month roadmap? What's going on? And what can you tell us around what's coming up?

SPEAKER_03 42:46

Yeah, I wish we had a 12-month roadmap. I think, you know, for the next few months and even the next few weeks, the biggest thing we are launching are specialists, which is Media Buying Specialists is now live. We have a CRO specialist that's actually generating landing pages for Shopify merchants and deploying them to Shopify. It's awesome. It allows you to iterate on landing pages, build landing pages for specific products, build full funnels. Now you can not only do the media buys, but also have the image ads and the landing pages all in one flow. Through the end of this year, it will be continuing to add into that. So, what are we adding into it? Well, we're adding MMM into it. So now Moby's got upper and lower bounds on budgets and can run simulations based on a bespoke model for you. He can run lift tests for you. Those things are all kind of coming live. And I think that is going to be a big, big focus for us through the back half of this year is how do we start to take all of the bits and pieces we brought together into, you know, one automated media buyer and really automated operator across the broader business. I don't want Angus to have to ask a question. I want Moby to be teeing up the answers. You know, here's here's what you need to do for fall of 26 in the UK. And this needs to be live, you know, next month because it's going to start getting cold at the end. August and September, and people are thinking fall. So the more we can take out even the proactive work that people need to do to prompt Moby is a big, big part of aggregating and understanding how people are building really cool stuff with it today as well.

SPEAKER_02 44:12

And I can see a beautiful, I think it's a beachside setting behind Angus. I'm sure he'd love to be out there doing more of that and letting Moby do the work inside. I think it sounds pretty good for everyone.

SPEAKER_00 44:21

Oh, it'd be amazing. Media buying is dead, I think they're saying. So press the button and come back later.

SPEAKER_02 44:27

But that is the theme that I'm picking up. Not that media buying is dead. Don't hang up on us yet, media buyers, but I think that is a theme that I'm picking up of how exciting it is right now in this space, because it seems like it wasn't that long ago that we were mourning the death of Google Universal Analytics. And we're like, how are we going to know anything because Universal's gone and I can't read GA4? And I was like, actually, we're not doing that. We're not trying to just understand dashboards and data. We're not even aiming to solve attribution perfectly. We're trying to automate a lot of the stuff that machines that are much smarter than us can make better decisions and much more rational decisions and automate that stuff and then provide the creativity on top of that and find those areas where businesses can be unique and get that edge rather than doing the stuff that was just reading the data and then turning that into action. So I think we're in a really exciting space. And to your point, Zach and your point, Angus, it's gonna look totally different again in 12 months. So I think we should have this chat again.

SPEAKER_03 45:29

That'll be fun. We can see all the video ads that Angus launched.

SPEAKER_02 45:33

That's it, exactly.

SPEAKER_03 45:34

I do think there's one important thing to note is I think strategy, taste, you know, brand, like those are the things that are have always won and I think will become a bigger focal point. And my other hope for brands is everyone's been so ingrained and entrenched in Meta and Google, and we love them. They're amazing partners, and people are spending a lot of money there, and they should. But my hope is that we free up the media buyers to start testing new channels to

Security Guardrails And The Near Future

SPEAKER_03 46:00

go and do something, you know, fun and crazy that you know marketers love to do. And if we can accomplish that, then the brands get bigger, the awareness gets bigger, the top of funnel starts to open up again. And you know, that's my hope that we can automate the pushing and pulling of levers and bring it back up to you know, brand taste, strategy, top of funnel, do TV buys again, you know, the things that are are really fun and exciting but take time.

SPEAKER_02 46:21

So is that what we should be getting out of this conversation? TV is the future. No, I love it. I I totally agree with you. The more we can allow brands to experiment and be in control of their brand destiny, the better, because I feel like we're at this point where there's a lot of same-same in e-commerce. I think Baird's a great example of going out with your own product and your own innovations and your own brand, and we need to enable more e-commerce brands and retail brands to be doing that. Thank you both for your time. Zach, I've got to ask you, you're obviously based in the US and you have an amazing triple whale team here in Australia. Most people will know the two atoms in market, as well as Alex and everyone else that's here now. If people are curious around triple whale and trying to do a little bit more research of whether it's right for them, where would you send them in the triple whale ecosystem?

SPEAKER_03 47:10

Yeah, I would go to triple whale.com. I mean, you're also welcome to DM me on X or LinkedIn. I'm happy to do a demo with you late night. I love getting on with the folks in Australia, and our team there is amazing, and I'm happy to join them on it. But yeah, triplewell.com's got everything you need. If you try to or are interested in a demo, it will get routed to the Australian team. So you'll have someone with plenty of overlap for you to support your time zone and support there and customer success there as well.

SPEAKER_02 47:35

Beautiful. Well, thank you so much, Zach. Thank you for your support, Triple Whale. We love having these conversations. And thank you so much, Angus, for giving us the real life examples of how you've used Triple Whale to make those business decisions and those performance marketing decisions. Appreciate it, Nathan. Zach.

unknown 47:52

Thank you.

SPEAKER_02 47:52

Cheers. I promised you a nerd out on all things data, measurement, and attribution.

Key Lessons And Closing Thanks

SPEAKER_02 48:00

And hopefully that's what we've delivered with a really great insight on Triple Whale, if it's a platform you're using or you are considering. So there are a few key lessons to come out of this one, as per normal. The first one is very simple. If you don't trust your data, fix that before you build anything on top of it. Angus told us that Baird spent about three months getting their numbers into a warehouse and checking them line by line before they even switched Triple Well on. Triple Well wasn't going to solve bad data. That upfront effort is exactly what let the whole team stop second guessing their data and start taking action quicker. Trust isn't a nice to have when it comes to e-commerce numbers. It's the thing that actually allows you to accelerate, to make decisions, to take action. And it needs to happen regardless of what platform you use. Secondly, next time your team is stuck in a debate, let the data settle it. Zach shared a brand that was arguing over a P-MAX campaign. The AI said kill it. They did, and it turned out to be the right call. Now, AI isn't always going to be right, but it's a pretty good starting point when it comes to people who can't agree. We all walk into decisions with an opinion we're either consciously or subconsciously attached to. There is no excuse now not to take opinion out of it and let the numbers and AI guide a decision. Doesn't mean it stops the conversation, but it will accelerate us past that initial argument. And the third thing out of this episode, once you trust the data, start handing the repetitive decisions to AI. But give it guardrails. True Classic runs their entire meta budget through Moby with no human in the loop, guided by a 10-page rules file that they wrote themselves. Zach himself says that they're ahead of the game, but that's where all of this is heading. Bed have built an autonomous buyer to keep scaling the US overnight while Australia sleeps. Do that well, and you free yourself up for the work that only you and your team can do. The things that differentiate you in markets, such as brand, customer service, creative. That is really exciting. As we heard in that episode, we are moving from understanding data to making decisions from data we can trust to automating actions from this data. It's a really exciting time in measurement and e-commerce, and we've got to think around how we actually put this data to use. Now, a big thank you for Triple Well for making this episode possible. We really do appreciate their support, and it's great to have them on board. If you got something from this one, please do subscribe or leave us a review. Until next time.